Package si.ijs.kt.clus.error

Class SpearmanRankCorrelationPerExample

java.lang.Object

si.ijs.kt.clus.error.common.ClusError

si.ijs.kt.clus.error.common.ClusNumericError

si.ijs.kt.clus.error.SpearmanRankCorrelationPerExample

All Implemented Interfaces:

Serializable

public class SpearmanRankCorrelationPerExample
extends ClusNumericError

This class computes the average spearman rank correlation over all target attributes. Ties are not taken into account. The spearman rank correlation is a measure for how well the rankings of the real values correspond to the rankings of the predicted values. This version of coefficient computes the correlation between multiple true/predicted target values for a given example, and returns the average over all examples. Some methods are missing.

Author:

beau, matejp

See Also:

• Serialized Form

Field Summary

Fields

Modifier and Type	Field	Description
protected ArrayList<Double>	RankCorrelations

Fields inherited from class si.ijs.kt.clus.error.common.ClusNumericError

m_Attrs

Fields inherited from class si.ijs.kt.clus.error.common.ClusError

DEFAULT_ERROR, DEFAULT_POSTFIX, DETAIL_ALWAYS_VISIBLE, DETAIL_NEVER_VISIBLE, DETAIL_SMALL, DETAIL_VERY_SMALL, m_Dim, m_Parent, RELATIVE_ERROR, RELATIVE_POSTFIX, TREE_ERROR, TREE_POSTFIX

Constructor Summary

Constructors

Constructor	Description
SpearmanRankCorrelationPerExample(ClusErrorList par, NumericAttrType[] num)
SpearmanRankCorrelationPerExample(ClusErrorList par, NumericAttrType[] num, String info)

Method Summary

Modifier and Type	Method	Description
void	addExample(double[] real, double[] predicted)
void	addExample(DataTuple real, DataTuple pred)
double	getAvgRankCorr()	Gives the average (=arithmetic mean) spearman rank correlation over all examples.
ClusError	getErrorClone(ClusErrorList par)
double	getHarmonicAvgRankCorr()	Gives the average (=arithmetic mean) spearman rank correlation over all examples.
double	getHarmonicRankCorrVariance()	Gives the variance of the harmonic mean of the rank correlation over all examples
double	getModelErrorComponent(int i)
String	getName()	Information
double	getRankCorrVariance()	Gives the variance of the arithmetic mean of the rank correlation over all examples
double	getSpearmanRankCorrelation(double[] a, double[] b)
boolean	shouldBeLow()	Should this error be minimized (e.g.

Methods inherited from class si.ijs.kt.clus.error.common.ClusNumericError

getAttr

Methods inherited from class si.ijs.kt.clus.error.common.ClusError

add, addExample, addInvalid, compute, computeLeafError, computeTreeErrorClusteringAbsolute, get_accuracy, get_auc, get_error_classif, get_precision, get_recall, getAdditionalInfo, getAdditionalInfoFormatted, getCoverage, getDetailLevel, getDimension, getErrorClone, getFormat, getModelError, getModelErrorAdditive, getModelErrorStandardError, getNbCover, getNbExamples, getNbTotal, getParent, getPrefix, isComputeForModel, isMultiLine, normalize, reset, setAdditionalInfo, setWeights, showDoubleArray, showDoubleArray, showDoubleArray, showModelError, showModelError, showRelativeModelError, updateFromGlobalMeasure

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Details

RankCorrelations

protected ArrayList<Double> RankCorrelations

Constructor Details

SpearmanRankCorrelationPerExample

public SpearmanRankCorrelationPerExample(ClusErrorList par,
 NumericAttrType[] num)

SpearmanRankCorrelationPerExample

public SpearmanRankCorrelationPerExample(ClusErrorList par,
 NumericAttrType[] num,
 String info)

Method Details

addExample

public void addExample(double[] real,
 double[] predicted)

Specified by:

addExample in class ClusNumericError

addExample

public void addExample(DataTuple real,
 DataTuple pred)

Overrides:

addExample in class ClusError

getModelErrorComponent

public double getModelErrorComponent(int i)

Overrides:

getModelErrorComponent in class ClusError

getAvgRankCorr

public double getAvgRankCorr()

Gives the average (=arithmetic mean) spearman rank correlation over all examples.

Returns:

average spearman rank correlation

getHarmonicAvgRankCorr

public double getHarmonicAvgRankCorr()

Gives the average (=arithmetic mean) spearman rank correlation over all examples.

Returns:

harmonic mean of spearman rank correlations for each example

getRankCorrVariance

public double getRankCorrVariance()

Gives the variance of the arithmetic mean of the rank correlation over all examples

Returns:

variance of the average rank correlation

getHarmonicRankCorrVariance

public double getHarmonicRankCorrVariance()

Gives the variance of the harmonic mean of the rank correlation over all examples

Returns:

variance of the average rank correlation

getName

public String getName()

Description copied from class: ClusError

Information

Specified by:

getName in class ClusError

getSpearmanRankCorrelation

public double getSpearmanRankCorrelation(double[] a,
 double[] b)

getErrorClone

public ClusError getErrorClone(ClusErrorList par)

Specified by:

getErrorClone in class ClusError

shouldBeLow

public boolean shouldBeLow()

Description copied from class: ClusError

Should this error be minimized (e.g. RRSE) or maximized (e.g. accuracy)? True by default.

Specified by:

shouldBeLow in class ClusError